Inferring Compositional Style in the Neo-plastic Paintings of Piet
Mondrian by Machine Learning
Overview
As part of the investigations described in [1], we developed software
for extracting formal representations of the Neo-plastic paintings of
Piet Mondrian. These representations simply enumerate the discrete
elements of the composition: line positions, rectangle locations and
colors, etc. This representation can be thought of as a set
of instructions for drawing a Mondrian painting, instead of
simply supplying a matrix of pixel color values (analogous to the
difference between vector and raster graphics). We created these
representations in order to facilitate computer analysis of the
compositional style of the paintings, and hope others may find this
data useful as well.
Data and code
This dataset contains formal representations for 46 of Mondrian's
Neo-plastic paintings, as well as 11 "earlier states" of Transatlantic
paintings reconstructed from the technical examinations found in [2].
The data are provided as a MATLAB *.mat file, and code for rendering
and saving images from the representations is provided.
MondrianData.tgz
Questions/Comments/Bugs
Open up your Python interpreter and e-mail me at:
'@'.join(['andrzeje','.'.join(['cs','wisc','edu'])])
Reference
[1]
Inferring Compositional Style in the Neo-plastic Paintings of Piet
Mondrian by Machine Learning
David Andrzejewski, David G. Stork, Xiaojin Zhu, and Ron Spronk.
Electronic Imaging: Computer Image Analysis in the Study of Art (SPIE 2010)
(pdf,slides)
[2]
Mondrian: The Transatlantic Paintings
Harry Cooper and Ron Spronk.
Published by Harvard Art Museum (2001)
